Bird Focus Accuracy Comparing SH2 with Normal Burst FPS Priority and with IS Priority.

[drj3]

The following results are for the OM1.2 with the MC20+300mm f4. However, I think there are implications for other lenses.

The test target was an image on my 27-inch (2550x1440) monitor. The bird image was 1.25 inches long and the monitor was photographed from an angle from 6.64 feet. The image depth of field is .12 inches, which makes it easy to see where the camera focused when viewing the darker inf focus monitor pixels on the image. Image #1 shows the angled monitor and Image #2 is one of the MC20+300mm images showing head focus.

Equipment & Settings. OM1.2+MC20+300mm f4. Release Priority = ON, Subject Detection = Bird, C AF Area = All (follows target outside focus area), C AF Sensitivity =+2. Three series of each condition (SH2, Normal Burst FPS Priority, Normal Burst IS Priority) where run in counterbalanced order. In the first series a single center focus point was used, in the final two series the default Large focus area was used. For the first series the focus point was placed on the bird’s eye. There was no difference in the results for the single and Large focus areas so the results were combined in the summary table.

Shutter preferred mode was used with a shutter speed of 1/160 (minimum for SH2). The frame rate for SH2 was set to 12.5 fps, for the normal burst modes the frame rate was set to 20. However, actual frame rate for normal burst modes is reduced by time required for focus and time required for stabilization in the IS Priority condition. The ISO was 6400.

I evaluated each image and classified images as Head focus if any part of the bird’s beak or eye was within the .12 inch depth of field. Images with focus behind the eye and in front feet were classified as Body focus. Images focused on the bird’s feet and tail were classified as Tail focus. Images in front of the bird, just outside the depth of field were classified as Small Focus Error and Large Focus Errors were more obviously outside the depth of field.

The IS Priority images showed the most difference in framing between image caused by the camera recentering the images for each frame and the SH2 images showed the least difference in framing between images. The actual FPS rates are shown in the table in Image #3. Unfortunately, there is no simple way to produce the same frame rates for the different conditions, since they will be different for different targets.















--
drj3

 

[mikero]

Very interesting. Thanks for posting. I can't see you mention it, but I assume the camera was on a tripod?

Mike

 

[HicHic]

73% of the head in focus on a still subject seems pretty close to what I get in practice. But when you have more pixels on target and can see whether the focus actually nailed the eye (as opposed to other parts of the head), the hit rate % of the eye is even lower than 73%. That number gets even lower when the subject is moving.

 

[drj3]

Very interesting. Thanks for posting. I can't see you mention it, but I assume the camera was on a tripod?

Mike

Not on a tripod. Hand held. I wanted to see if the recentering of the image and image movement between frames hand holding the camera caused a shift in the focus area on target. This does mean that different users could get different results depending on differences in how well they kept the target in the frame.

What surprised me was that the IS Priority stayed on its starting focus position better than the FPS modes where the framing changed much less between frames.

When I get time, I may try the test on a 3-dimentional target to see if I get a similar result. Using the angled computer monitor makes it extremely easy to see where the camera focused, but a real target could give a different result.

 

[drj3]

73% of the head in focus on a still subject seems pretty close to what I get in practice. But when you have more pixels on target and can see whether the focus actually nailed the eye (as opposed to other parts of the head), the hit rate % of the eye is even lower than 73%. That number gets even lower when the subject is moving.

I am sure the results would be somewhat different for different targets and target position. The setting of C-AF Sensitivity may also affect where the camera focuses as the moves, especially if there are wings which change distance as the bird moves wing positions and there is no clear eye.

The greater variability of the focus location with SH2 surprised me. My previous tests of SH2 vs normal burst indicate that SH2 needs a little more detail change than normal burst mode to change focus. In this test the only movement and detail change would be my movement hand holding the camera, which I would expect to be less with SH2 than when using modes with blackout.

Personally, I haven't noticed a real problem with focus location with the OM1s. My result gives me one possible reason for that.

I have two Custom mode settings for BIFs. Both use a large focus area with Shutter Priority set to 1/2000, which I either raise or lower depending on the amount of light. One is for SH2. the other is for normal burst with IS Priority. If the light is sufficient for faster shutter speeds, I choose the custom SH2 mode. If the light requires a shutter speed of 1/2000 or slower, I generally choose IS Priority mode. Using the MC20 on the 300mm and the minimum f8 aperture means, I use the IS Priority mode more than SH2.

 

[mikero]

That's an interesting result. I'd assumed that the more consistent framing shown with FPS Priority would lead to better maintaining the focus on the intended point on the subject. It appears not so according to your test. So if you can accept the lower frame rate, then IS Priority is the better option.

Would be nice to see the test repeated with a 3D subject if you have time.

Mike

 

[David5833]

Your excellent idea of looking at the monitor pixels to locate the point of focus is something that would be hard to mimic with a 3-D subject at different birding distances, but probably something worthwhile to compare to eliminate the possible influence of monitor screen refresh rates. You did a lot of work and it does beg several questions, but after looking at almost 1000 images, you likely don't want to make this a research career .

One thing that stands out to me is that focus accuracy seems to be strongly negatively correlated with frame rate, but of course it might not be that simple over a wider range of frame rates.

--
https://www.flickr.com/photos/143821723@N06/

 

[drj3]

Your excellent idea of looking at the monitor pixels to locate the point of focus is something that would be hard to mimic with a 3-D subject at different birding distances, but probably something worthwhile to compare to eliminate the possible influence of monitor screen refresh rates. You did a lot of work and it does beg several questions, but after looking at almost 1000 images, you likely don't want to make this a research career .

One thing that stands out to me is that focus accuracy seems to be strongly negatively correlated with frame rate, but of course it might not be that simple over a wider range of frame rates.

I think the frame rate may have been the most important variable. The MC20+300mm is slower to focus than most lenses without TCs. That is the reason I included actual frame rate in the table. While I have not yet had time to try this without my TCs and various targets, I will try to do so in the future comparing the 300mm f4 without TCs on a target where it gives a frame rate comparable to the SH2 rate. Unfortunately, there is no way to include the IS Priority in that comparison, since it tops out at 7-8 fps with the 300mm f4 without TCs.

If this is true, then the OM1s are different from the E-M1s. The E-M1s used previous frames to predict target movement (according to Olympus) as well as current focus and frame rate had little effect on percentage of well-focused images. The OM1s may be only using their ability to quickly focus and not be using previous frames for predicting target distance. SH2 at high frame rates may still give more correctly focused images along with a lot more images not well focused compared to lower frame rates with burst modes. However, most frames with all three modes had some part of the target in focus. This would in many situations give good images with most of the target in focus.

--
drj3

 

[Skeeterbytes]

Very interesting. Thanks for posting. I can't see you mention it, but I assume the camera was on a tripod?

Mike

Not on a tripod. Hand held. I wanted to see if the recentering of the image and image movement between frames hand holding the camera caused a shift in the focus area on target. This does mean that different users could get different results depending on differences in how well they kept the target in the frame.

What surprised me was that the IS Priority stayed on its starting focus position better than the FPS modes where the framing changed much less between frames.

When I get time, I may try the test on a 3-dimentional target to see if I get a similar result. Using the angled computer monitor makes it extremely easy to see where the camera focused, but a real target could give a different result.

I'd offer my pet eagle but the FedEx charges are ghoulish.

Interesting outcomes. It's not often I wring out every settings permutation and this is a good case for doing just that. Heck, sometimes I can't even count them.

Rick

 

[HicHic]

73% of the head in focus on a still subject seems pretty close to what I get in practice. But when you have more pixels on target and can see whether the focus actually nailed the eye (as opposed to other parts of the head), the hit rate % of the eye is even lower than 73%. That number gets even lower when the subject is moving.

I am sure the results would be somewhat different for different targets and target position. The setting of C-AF Sensitivity may also affect where the camera focuses as the moves, especially if there are wings which change distance as the bird moves wing positions and there is no clear eye.

The greater variability of the focus location with SH2 surprised me. My previous tests of SH2 vs normal burst indicate that SH2 needs a little more detail change than normal burst mode to change focus. In this test the only movement and detail change would be my movement hand holding the camera, which I would expect to be less with SH2 than when using modes with blackout.

Personally, I haven't noticed a real problem with focus location with the OM1s. My result gives me one possible reason for that.

I have two Custom mode settings for BIFs. Both use a large focus area with Shutter Priority set to 1/2000, which I either raise or lower depending on the amount of light. One is for SH2. the other is for normal burst with IS Priority. If the light is sufficient for faster shutter speeds, I choose the custom SH2 mode. If the light requires a shutter speed of 1/2000 or slower, I generally choose IS Priority mode. Using the MC20 on the 300mm and the minimum f8 aperture means, I use the IS Priority mode more than SH2.

Yeah for sure. If you have wing movement and more dynamic changes in the bird's shape and size, then it will make it even more challenging for the camera.

 

[daveomd]

Very interesting, thanks for doing the test.

The test caused a few questions:

a. Would the MC 14 make a difference vs the MC 20? ( I use my 300mm f4 with the mc 14)

b. Would using 25fps vs 12.5 fps? ( I shoot Peregrine Falcons which are incredibly fast)

c. Does Sensitivity matter? I use +2, would 0 or -2 be better?

The variables in settings on the OM-1.2 are amazing. Thanks in advance.

 

[drj3]

Very interesting, thanks for doing the test.

The test caused a few questions:

a. Would the MC 14 make a difference vs the MC 20? ( I use my 300mm f4 with the mc 14)

b. Would using 25fps vs 12.5 fps? ( I shoot Peregrine Falcons which are incredibly fast)

c. Does Sensitivity matter? I use +2, would 0 or -2 be better?

The variables in settings on the OM-1.2 are amazing. Thanks in advance.

I haven't yet tested the MC14 or the lens without TCs. However, since the cameras focus speed definitely slows down as the aperture size decreases, I would expect performance to be better with the MC14 and even better with the lens alone. Of course, focus accuracy also depends on the illumination and detail of the specific target.

Frame rate probably has an effect on focus accuracy on fast flying birds, but probably less important with fast apertures like f2.8 or f4. The OM Systems warning focus accuracy for apertures smaller than f8 is specifically for the faster Sh2 and ProCapSh2 where it is likely to result in more unfocused or poorly focused images.

I did some tests with the OM1.1 on its original firmware and normal bursts changed focus more quickly than Sh2 at 25fps and SH2 at 25 fps changed focus more quickly than 50 fps. Later firmware updates reduced the differences between the three conditions.

Sensitivity +2 will be more like to cause a quick change in focus if the camera miss identifies the target but is more likely to retain focus on a target that changes speed/distance quickly. We do not know specifically what the Sensitivity settings do. I saw some tests on the original E-M1.1 which stated that it was a variable brief hold time which limited the amount of change in focus distance allowed.